This proposal is a bit different from a concrete topic, but it's based
on the generally positive feedback from the end user panel last year.
It's to allocate 2h of Kernel summit to discuss problems raised by
vendors and end users. The format I'm proposing is 10 minutes of
problem introduction by someone from the vendor or user representative
who is a subject matter expert and 20 minutes of discussion (so,
hopefully four such problems).
As usual, the difficulty is actually finding people from the vendor and
user community willing to do such a presentation (and ensuring they
won't simply want to talk about their problems with X distro enterprise
kernels). The way I think we might get some diverse problems is to send
out a call for participation. Something like.
"The kernel summit programme committe is inviting proposals for
participation in the 2008 Kernel Summit in Portland Oregon (URL) from
members of the End User and Vendor communities wishing to raise
particular issues or problems they see with the current Linux Kernel. A
good problem presentation should stimulate discussion with the kernel
community, and as such the presenter should be a subject matter expert
capable of engaging in the debate. While the presenter may have a
preferred solution, they should concentrate on the problem and be
prepared to consider the effects of solutions outside of their preferred
one. It goes without saying that problems should have a significant
component which is susceptible to a solution within the Linux Kernel.
"Proposals of around 500 words and describing the nature of the problem
and the credentials of the presenter should be forwarded to the Linux
Kernel Programme Committee at
<[email protected]>"
Obviously we'd have to circulate this beyond the usual mailing lists,
but we can solicit help from the Linux Foundation to do this.
James
On Fri, Jun 27, 2008 at 9:21 AM, James Bottomley
<[email protected]> wrote:
> This proposal is a bit different from a concrete topic, but it's based
> on the generally positive feedback from the end user panel last year.
> It's to allocate 2h of Kernel summit to discuss problems raised by
> vendors and end users. The format I'm proposing is 10 minutes of
> problem introduction by someone from the vendor or user representative
> who is a subject matter expert and 20 minutes of discussion (so,
> hopefully four such problems).
>
> As usual, the difficulty is actually finding people from the vendor and
> user community willing to do such a presentation (and ensuring they
> won't simply want to talk about their problems with X distro enterprise
> kernels). The way I think we might get some diverse problems is to send
> out a call for participation.
I think this is a good idea and I suggest using referrals since it's not
just about the problem. It's also how it's presented. And it would be
worth a few $K to pay for a good presenter who has something relevant
to say. e.g. enough to cover travel/hotel plus something extra.
Secondly, these problems arise because second tier users aren't
participating in the development. I see this CFP as a "Call For Participation"
and not a "call for papers".
I also suggest trying to get representatives from different communities.
e.g. education, handhelds, desktop (something other than gfx drivers,
maybe revisit the user visible latency issues), HPC, database, etc.
I left out "embedded" since that's now too vague, has been flogged
to death in the past, and I _hope_ now has better participation.
The last issue I see is "Vendor" usually means HW vendor with a few
major SW vendors (e.g. Oracle, SAP). Those guys have sufficient
representation and I think some newer, smaller players (e.g. not google)
would be more interesting to invite. Maybe dreamworks, pixar, or
Ernie Ball Guitar Strings (http://www.ernieball.com/) might be examples.
hth,
grant
> Something like.
>
> "The kernel summit programme committe is inviting proposals for
> participation in the 2008 Kernel Summit in Portland Oregon (URL) from
> members of the End User and Vendor communities wishing to raise
> particular issues or problems they see with the current Linux Kernel. A
> good problem presentation should stimulate discussion with the kernel
> community, and as such the presenter should be a subject matter expert
> capable of engaging in the debate. While the presenter may have a
> preferred solution, they should concentrate on the problem and be
> prepared to consider the effects of solutions outside of their preferred
> one. It goes without saying that problems should have a significant
> component which is susceptible to a solution within the Linux Kernel.
>
> "Proposals of around 500 words and describing the nature of the problem
> and the credentials of the presenter should be forwarded to the Linux
> Kernel Programme Committee at
> <[email protected]>"
>
> Obviously we'd have to circulate this beyond the usual mailing lists,
> but we can solicit help from the Linux Foundation to do this.
>
> James
>
>
> _______________________________________________
> Ksummit-2008-discuss mailing list
> [email protected]
> https://lists.linux-foundation.org/mailman/listinfo/ksummit-2008-discuss
>
On Fri, 2008-06-27 at 11:21 -0500, James Bottomley wrote:
> This proposal is a bit different from a concrete topic, but it's based
> on the generally positive feedback from the end user panel last year.
> It's to allocate 2h of Kernel summit to discuss problems raised by
> vendors and end users. The format I'm proposing is 10 minutes of
> problem introduction by someone from the vendor or user representative
> who is a subject matter expert and 20 minutes of discussion (so,
> hopefully four such problems).
>
> As usual, the difficulty is actually finding people from the vendor and
> user community willing to do such a presentation (and ensuring they
> won't simply want to talk about their problems with X distro enterprise
> kernels). The way I think we might get some diverse problems is to send
> out a call for participation. Something like.
>
> "The kernel summit programme committe is inviting proposals for
> participation in the 2008 Kernel Summit in Portland Oregon (URL) from
> members of the End User and Vendor communities wishing to raise
> particular issues or problems they see with the current Linux Kernel. A
> good problem presentation should stimulate discussion with the kernel
> community, and as such the presenter should be a subject matter expert
> capable of engaging in the debate. While the presenter may have a
> preferred solution, they should concentrate on the problem and be
> prepared to consider the effects of solutions outside of their preferred
> one. It goes without saying that problems should have a significant
> component which is susceptible to a solution within the Linux Kernel.
>
> "Proposals of around 500 words and describing the nature of the problem
> and the credentials of the presenter should be forwarded to the Linux
> Kernel Programme Committee at
> <[email protected]>"
>
> Obviously we'd have to circulate this beyond the usual mailing lists,
> but we can solicit help from the Linux Foundation to do this.
Since there wasn't much feedback on this, the programme committee
thought it would be unlikely to produce much in the way of proposals, so
instead, we'll do what we did last year: recruit two or three end users
to do a lightning type talk followed by 20 minutes or so of discussion.
For those of you who enjoyed the end user presentations last year and
wanted to go into greater depth than we can in a single KS session, the
Linux Foundation is holding an end user summit in October in New York:
http://www.linuxfoundation.org/events/enduser
Since it's already in Manhattan, and a lot of the major banks and other
end users with enterprise class problems are attending, we're going to
try to schedule individual sessions with some of their architects and
kernel developers to go over some of the interesting problem cases.
KS attendees who want to come and whose employers won't pay, we can
likely sponsor (at least some) from the LF community travel fund:
http://www.linuxfoundation.org/travel/
James